World-State
World-State is the product interface between video and software.
Raw video is expensive to search repeatedly. World-State turns footage into structured objects, events, zones, timestamps, relationships, confidence, and evidence references that software can inspect.
World-State data model
| Part | Purpose | Example fields |
|---|---|---|
| Objects | Represent things observed over time | object_id, label, confidence, first_seen, last_seen |
| Events | Represent changes or actions | event_type, timestamp, participants, evidence_refs |
| Zones | Represent spatial context | zone_id, name, geometry, camera_ref |
| Relationships | Connect objects and events | subject, relation, object, confidence |
| Review flags | Expose uncertainty | low_confidence, ambiguous_scene, human_review_required |
Why this matters to product teams
World-State lets teams build dashboards, search, APIs, alerts, and agents over a stable representation. It does not make every visual problem solved. It gives the product a durable memory layer that can be tested, versioned, cached, and compared against evidence. That is what makes repeated queries cheaper and debugging easier.
World-State quality checklist
- Important objects appear with stable IDs.
- Events include timestamps and evidence_refs.
- Low-confidence observations are not hidden.
- Ambiguous events become review_required.
- Updates are versioned so agents know what state they read.
Product boundaries are part of the product, not footnotes. Ayneye is not presented as a replacement for every GPU video foundation model, not a free-infinite-query engine, and not an autonomous surveillance decision system. The current beta path is controlled signup, dashboard, REST API, bounded processing, evidence artifacts, visible limits, and read-only agent/MCP-style evaluation. Hard scenes, identity-sensitive workflows, emergency response, physical access, discipline, and destructive actions require human review or remain blocked.